Font Size: a A A

On Cooperative Localization And Sink-location Privacy Of Multi-agent Systems

Posted on:2016-08-20Degree:DoctorType:Dissertation
Country:ChinaCandidate:G F ChaiFull Text:PDF
GTID:1108330482973773Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Due to the spatial and functional distribution, strong fault tolerance, robustness as well as scalability, multi-agent systems (e.g. unmanned vehicles and mobile sensors) are widely used in military, industrial and agricultural production, space exploration, medical, traffic control, ser-vice industry and other fields. Localization is a basic problem of multi-agent systems, since the knowledge of location information is usually the premise for multi-agent systems to perform com-plex tasks. Meanwhile, the location privacy of sensor networks has been a important security prob-lem. Sink nodes are the aggregation center of sensing data and vital to ensure the normal function of sensor networks. Once the location of sink nodes are exposed to enemies and sink nodes are destroyed, it will not be able to send sensing data to data centers and the total network will run out. However, the open nature of wireless communication media makes the protection of location privacy a difficulty.This paper addresses the localization problem of sources and agents in mobile multi-agent sys-tems in different settings. Different distributed observers are proposed to dynamically estimate the location of sources or agents and the conditions under which the observers converge are provided. It is the first time that k-anonymity is introduced to solve location privacy protection problem of sink nodes. A routing protocol based on Euclidean minimum spanning tree has been provided in order to balance two conflicting goals which are location privacy and the routing energy cost. The main work and results of thesis can be cataloged as follows:Firstly, based on range measurements, the cooperative source localization problem are ad-dressed in a continuous time framework. The goal is to estimate the coordinate of a stationary source for each agent in its own local frame. Suppose the measurements and exchange of infor-mation between agents are bidirectional, so the topology of the network is an undirected graph. a localization algorithm based on linear time-varying differentiator using single mobile agent and a consensus-based fusion scheme are proposed. The asymptotical convergence of the algorithm is proved by using persistent exciting condition and input-to-state stability theory. The require- ments of the topology of the network and the movement of the agents are also given to ensure the convergence.Secondly, the cooperative source localization problem are addressed in a continuous time framework based on sampled range measurements. Measurements and information exchange be-tween neighboring agents are are supposed to be directional, so the network topology is a directional graph. First, we propose a discrete-time source localization algorithm using single mobile agent. The convergence of the algorithm is proved by the construction of a Lyapunov function. Also the requirements on the sampling period and the movement of the agent are given to ensure the con-vergence.Then, a consensus-based estimation fusion scheme is proposed to fuse the estimates from neighbor agents. A sufficient condition under which the algorithm converges is provided and the convergence is proved via discrete-time input-to-state stability theory.Next, self-localization problem of mobile multi-agent systems is addressed. The measure-ments and information exchange between neighboring agents are supposed to be directional and directional, so the network topology is a time-varying directed graph. A continuous-time observ-er with state switches is proposed and the asymptotical convergence of the observer is proved via persistent exciting condition, input-to-state stability theory and the stability theory of cascade sys-tems.The requirements on the network topology and the movement of the agents are given to ensure the convergence.Finally, the sink location privacy of wireless sensor networks is addressed. To defend against powerful global adversaries, the k-anonymity scheme is introduced to preserve sink location priva-cy for the first time. To achieve a balance between the location privacy and the energy consump-tion of the system, a routing protocol based on Euclidean minimum spanning tree is proposed. Therefore, the problem is transferred to a complex nonlinear optimization problem with nonlinear constraints. Two efficient algorithms are provided to resolve the optimization problem.This dissertation gives rigorous theoretical proof for the proposed analysis and control design methods. Several simulations are given in each chapter to illustrate the results.
Keywords/Search Tags:Multi-agent, localization, consensus, sensor network, sink node, location privacy, k- anonymity
PDF Full Text Request
Related items